Thom Hickey wrote:
Ralph LeVan here at OCLC has worked on an SRU interface to Lucene.
The combined indexer/SRU approach is the tack I've been taking regarding search too. I don't care a whole lot whether I use this indexer, that indexer, or the other indexer as long as I can make sure I have an SRU, OpenURL, Z39.50, etc. interface to the index. This will always allow me to swap out the an older indexer for a new one as they become available. BTW, the SRU Perl module on CPAN work with Lucene quite nicely, as well as swish-e, Plucene, and Kinosearch. Through the Perl ZOOM module I will be able to use Zebra as well with the Perl SRU module. When it comes to a storage mechanism, I flip-flop between saving things to a database and then writing reports against the database feeding them to the indexer, or using only an indexer. Furthermore, I flip-flop between using a native XML database to contain my content or a relational one. I know there are advantages and disadvantages of all the choices, but at least it is good to have choice. -- Eric Morgan